sun5.9 coredumps
 

Hello,

Are there any root(user) log file in Solaris5.9 os version get populated if there any core dump happens on Solaris. If so which file is that in Solaris?

I am facing core dump on Solaris, I would like to see what is the real issue happened when it coredumps by seeing some log file in /root area.

Please let me know.

Thanks & Regards
Dayalan

Have a look at the coreadm manual page.

Apart from 'coreadm' you might also try 'appcrash'.
http://blogs.sun.com/gregns/

if you are talking about the system crashing and not an app, then type dumpadm and not coreadm. dumpadm will tell you if dumps are enabled and where they are stored (/var/crash/machine_name by default).

You can use mdb to analyze them

ex:

cd /var/crash/mybox

mdb -k unix.0 vmcore.0

::status
$<msgbuf
$q

If you are talking about apps dying on you, run find / -name core and use mdb to analyze them as well.

sun5.9 coredumps
 

Hi javier,

I am not sure whether the machine or application crashing, How to debug this issue using mdb? i dont have much experience with the mdb, it looks like difficult for me to debug under mdb, reason is, it displays assembly level instruction when i am debugging under mdb. so it makes little bit difficult for me to narrow down the issue.

here is the obsveration about the issue.

1. I am not able to re-produce the application core dump always, some times it coredump (90%) and sometimes it doesnt core dump (10%).
2. we have three SUN 5.9 boxes, only one box in SUN5.9 out of three boxes, the application crashes and it displays in free() it crashes in the stack trace. but based on my analysis it free(ptr) is not really corrupting the heap instead somewhere before itself it (i am not sure whether the app or machine) corrupted the heap so i get the crash on free(ptr) in someother code in the same application.
3. As i said earlier the app crashes on only one sun5.9 box out of three sun5.9 boxes under any debugger (dbx, purify, watchmalloc on sun5.9) i am not able to re-produce the coredump in the same box (where it crashed).


why it crashes in one machine, why not it crashing on other two sun5.9 boxes.
why it crashes when i invoke from the shell prompt (in app crashed sun5.9 box) but it is not crashing under any debugger in the same box (where it crashed). Any inputs on this?

Please let me know your inputs.

Thanks & Regards
Deenadayalan

Hi:

If I understand this correctly, you have one application that dies occasionally in one machine out of three. So it is not the system that panics but the app that dies by itself.

If the app dies in one machine and not the others, check out for differences in configuration files or patches. Else, perhaps you have less resources on that particular machine (less memory, swap and so on).

Have you looked at /var/adm/messages at all or have you called the company that makes the app?

Just a few ideas...

J.

Hello,

I used dbx to find the core dump issue using "check -all" command, It crashed in RTC enabling stage itself, Can anyone help me? Can you please suggest me whether i am using correct dbx.

================================================================================================
Reading ld.so.1
Reading librtc.so
Reading libnsl.so.1
Reading libsocket.so.1
Reading libdl.so.1
Reading libswt03064.so
Reading libm.so.1
Reading libc.so.1
Reading libmp.so.2
Reading libc_psr.so.1
access checking - ON
Running: simdata_12
(process id 13212)
RTC: Enabling Error Checking...
dbx: internal warning: rtc: ld/st instruction uses %r6 at 0x10002d42c in `simdata`Ttop_1_1_0
dbx: system error: cannot recover; Access checking disabled
dbx: internal warning: rtc_new_process failed.
dbx: Error checking initialization failed. All error checking disabled.
==================================================================================================== ==========

Thank you in advance, Please let me know your inputs.

Thanks & Regards
Deenadayalan

sun5.9 coredumps
 

Hello,

I attached process id with dbx using (dbx - <process_id>) command, to find out the memory bugs, I enabled "check -all" in dbx but it generates error message follows.

=========================================================
(dbx) check -all
dbx: check will not work with attached process,if librtc is not preloaded.
See `help rtc attach'
=========================================================

I wanted to enable "rtc" but it gives above error, Please let me know your inputs.

Thanks & Regards
Deenadayalan